Cluster hierarchical vs. divisive strategy

by | Mar 17, 2018 | Machine learning | 0 comments

The hierarchical methods try to create groups of homogeneous elements among themselves and heterogeneous among groups, to achieve it mainly can be done by means of agglomerative or divisive strategy.

Example of item distribution for clustering

Example of item distribution

Agglomerative strategy

The strategies are based on a set of individual elements and together the elements that are most similar to the number of clusters that are considered optimal.

The computational complexity of these methods is very high in the order of N ^ 3, although there are algorithms that have reduced this complexity as will be seen later.

Agglomeration Cluster

Agglomeration Cluster

Divisive strategy

The divisive strategies depart from the set of complete elements and separate the groups that are more different between them until they remain with a number of clusters that are considered optimal.

The computational complexity of these methods is very high in the order of 2 ^ n, although there are algorithms that have reduced this complexity as will be seen later.

Divisive Cluster

Divisive Cluster

0 Comments

Submit a Comment

Your email address will not be published.